Do you want to be like Kuro5hin? Because this is how you get to be like Kuro5hin. 2016-08-26T15:32:52.823Z


Comment by Dentin on [deleted post] 2021-07-17T17:24:05.207Z

I find his post incredibly uncompelling.  I believe he's arguing against a straw man; economists in the real world doing real work involving real dollars are painfully aware of the issues he brings up.

My guess is that his experience in economics is heavily influenced by his PhD work and that he's arguing against "economics as he experienced it at universities" as opposed to "economics as practiced by professionals in a real economy".

Comment by Dentin on Up-to-date advice about what to do upon getting COVID? · 2021-06-20T17:23:49.559Z · LW · GW is one of the better, current overviews of Ivermectin.  Basically, we don't have enough information.

Comment by Dentin on Covid vaccine safety: how correct are these allegations? · 2021-06-14T20:41:56.871Z · LW · GW

One other thing to consider is that even small differences in replication rate might actually matter.  Consider that it takes a week for the virus to really ramp up, and that's a large number of doubling periods.  Even just getting a larger or smaller initial dose seems linked to how sick people get.  Even a few percent difference may allow the immune system to stay ahead in the arms race, and result in a nonlinear change in death rate.

Note that I'm not saying this happens; I'm saying that because this is an exponential growth attacker (the virus) versus and exponential growth responder (the immune system), even small differences in growth rates might have a large impact.

Comment by Dentin on Looking for reasoned discussion on Geert Vanden Bossche's ideas? · 2021-06-06T23:33:21.889Z · LW · GW

Yeah, his second claim is bogus.  That's not how it works, and that's not what we've been seeing with existing mutations.

As an example, look at E484K - this mutation changes the amino acid polarity, so that antibodies trained against the E variant will have a much harder time attaching to the K variant.  If an antibody fails to attach, it doesn't 'crowd out' anything.

In the case where an antibody attaches but doesn't actually "inactivate" the virus due to a mutation, that's because the virus' attack surfaces are still present and exposed (otherwise, it would be inactivated.)  Again, we wouldn't expect to see "crowding out" of other antibodies.

And lastly, there's the extremely unlikely scenario of sufficient mutation that existing antibodies give us Vaccine Enhanced Disease. This is both something vaccine designers explicitly focus on to minimize the risk of, and would require an extreme amount of change to enable.

Comment by Dentin on Looking for reasoned discussion on Geert Vanden Bossche's ideas? · 2021-06-06T20:22:54.607Z · LW · GW

#1 is where I would hinge a lot of objection.  Specifically:

"The vaccines are targeting outdated variants, and some vaccines are already only partly efficient. This creates the perfect conditions for further viral evolution."

Yes, the vaccines are targeting outdated variants, and yes, the vaccines are only partially efficient.  But the mRNA vaccines, even partially efficient, are still hugely overkill.  From previous posts here on LW, even partially effective mRNA vaccines likely cut transmission by a factor of 100 between the reduced infection rate and reduced infectiousness when infected.

On the other hand, "perfect conditions for viral evolution" require a much, much weaker vaccine response, one that barely keeps up with the spread of the virus.  For maximum evolutionary pressure, you'd want a continuous rate of infection, with a spreading factor very near 1.0, so even slight changes in the virus can be strongly selected for.  And it just so happens that this is exactly what we were doing prior to vaccine rollout.

Of course, even the mRNA vaccines aren't a guarantee that we'll eradicate the virus.  There's a handful of marginally effective vaccines out there (Sinovac anyone?) which will be much easier for mutations to overcome, and there's areas which will be infection hotbeds for years, and there's antivax communities which will likely be infection centers forever. But by broadly rolling out strong vaccines, there's every reason to believe that the 'nightmare scenario' of Bossche will be less likely, not more.

Comment by Dentin on Looking for reasoned discussion on Geert Vanden Bossche's ideas? · 2021-06-06T18:51:25.312Z · LW · GW

The core issues I had with it are that the scenario he's envisioning just isn't playing out in the real world.  He has a mostly coherent model he's working with, and he's making the valid claim that "for some set of parameters and constants, terrible things happen!"  It's just that in real life, the parameters and constants are nowhere near what's needed for the badness he's claiming.


  • His scenario requires a much higher mutation rate than we're seeing;
  • His scenario assumes vaccine immunity isn't very strong, but what we're seeing is incredibly strong immune responses;
  • His scenario assumes that natural immunity (from being infected) is significantly stronger than the vaccine generated immunity, which doesn't appear to be the case.

Basically, none of his assumptions are valid, and so his final prediction is for a reality that isn't ours.

Comment by Dentin on Why I Work on Ads · 2021-05-06T22:30:38.727Z · LW · GW

These extremely short responses discarding the bulk of my content feel less like you're attempting to understand, and more like you're attempting to get me to draw bright lines on a space I have repeatedly indicated is many different shades of grey.  Disconnecting from the discussion for now.

Comment by Dentin on Why I Work on Ads · 2021-05-06T22:24:53.079Z · LW · GW

Publisher, advertiser, the distinction does not matter.  The point is that the target does not get to decide.

Comment by Dentin on Why I Work on Ads · 2021-05-06T14:18:23.593Z · LW · GW

You might find it unpleasant, but it's it the job of Simurgh Followers to spread the Truth Of The Endbringers to everyone!  Surely if people just watch enough of it, they will be converted.

The point is that the target gets to decide what's acceptable and what isn't, not the advertiser. The current system makes the advertiser the judge, and that's not ok, even if we have managed to construct a sorta functional system that mostly takes care of the worst abuses.

Comment by Dentin on Why I Work on Ads · 2021-05-06T14:16:22.017Z · LW · GW

I'm not convinced I fully understand your distinction, let alone that we could codify it sufficiently to make it into law.

Regarding 'codify into law', that's not an excuse, and it disregards how the US legal system works.  If we can codify slander, if we can codify "harm", if current advertising companies can codify "unacceptable ad", we can codify this.

If you visit a model railroading site, are ads for model locomotives push or pull?

Firm push, but only because of the physical realities of the current system.

The fact of the matter is that by default, visiting a site isn't a directed action.  Clicking on links may take you anywhere, and links may be obfuscated.  My preference would be that any/all landing pages should be clean, and ads only shown for explicit searches requesting explicit content.  As a second best, I'd take 'only show ads on explicit navigation after page landing'.

Comment by Dentin on Fractal Conversations vs Holistic Response · 2021-05-06T11:57:52.557Z · LW · GW

For bulk exchange of information and state, holistic is really good.  I strongly prefer the holistic approach, but I've found that it only works for entirely friendly conversations.  If it's adversarial, I find that branches get aggressively pruned to just the things that the opposing side can most easily attack.

And if you think about holistic being optimized for "exchange of information and state", this makes perfect sense:  adversarial conversations are rarely if ever about information exchange; they're about "winning".

It's also perhaps worth mentioning that aggressively holistic comms can require more mental horsepower than some people have readily available or are willing to invest, so it's best to tune the level of threading based on audience and discussion type.

Comment by Dentin on Let's Go Back To Normal · 2021-05-06T11:54:45.733Z · LW · GW

Given the current vaccination rates in the US, and the fact that supply is already beginning to exceed demand, I'd recommend full open in approximately a month.  That gives most of the remaining unvaccinated people time to go get at least their first shot, and should allow us to get below exponential growth nationwide, even though we're likely to have it in sub-populations.  IMO the target should be 'not overloading hospitals'.

After that, let it burn.

From the interviews and things I've seen so far, literally the only way to change a vaccine denier's mind is for them to either 1) get sick, or 2) have a close contact get sick.  At this point, I'm kinda ok with that, if that's what it takes to get people's attention back to ground truth, instead of socially constructed fake reality.

I'm personally trying to push everyone I care about in the vaccine direction, and I would be terribly sad to lose one of my family members when I'm unsuccessful.  However, I also don't want to live in a future world where it's ok to be intentionally ignorant on decisions that affect your life.  If that's what it takes for my family to get a reminder that conspiracy theories are Not OK, that's what it takes.

Comment by Dentin on Let's Go Back To Normal · 2021-05-06T11:43:58.119Z · LW · GW

I'd just like to point out that while "facing these tradeoffs is stupid and avoidable" (which I agree with), it's much, much more accurate to say instead "facing these tradeoffs is effectively impossible to avoid even though it's stupid and avoidable".  We might not like reality, but it's not going to go away no matter how much we call it stupid and avoidable.

Comment by Dentin on Alzheimer's, Huntington's and Mitochondria Part 3: Predictions and Retrospective · 2021-05-06T11:40:10.524Z · LW · GW

Having an overarching model (or several competing models) of which different parts can be tested independently seems like a structure which is very amenable towards different scientists, so I am disappointed none of the biological/medical community has started doing something like this.

This is actually done, quite a lot in fact, it's just really hard and the search space is huge.  Kudos to you for your analysis; it's unlikely to be a major step forward, but given that idea search space is effectively exponential, it's also entirely possible that it's a unique insight.  Please do attempt to publish it.

For an 'off the top of my head' example of this sort of modeling in the wild, this is a really interesting paper:

Basically, the model was "we've got a class of cancer cells with mitochondrial weirdness, what happens if we shut off both mitochondrial ribosomes at the same time?"  And it turned out there were commonly available low dose drugs which do this.

Comment by Dentin on Why I Work on Ads · 2021-05-06T11:22:52.143Z · LW · GW

> its entire purpose is to alter people's mental state without their permission

I think that's the core of our disagreement?

Yes, and I think that would be a better path to attack my position.  There's two attack vectors in that quoted line - "alter peoples mental state without their permission", and "permission".  I would recommend avoiding the first attack vector; that will be an exceedingly difficult sell to me.

Permission on the other hand is already a partially open attack vector, and you're much, much more likely to change my mind by that route.  Examples:

  • I have very little objection to the ads on the Google web search interface.  I don't notice them, but I do sometimes click on them.  The reason I have no objection is because "I was actively looking for something", and the ads are almost always topical and don't drown out real results. In other words, I gave implicit permission by searching for the thing that was being advertised to me.
  • I have very little objection to the ads within the Amazon search interface.  Again, it's because I was explicitly looking for the thing in question, and typically the ads presented are factual results that answer my query.

In both those cases, permission isn't some implicit, distant concept; I explicitly make the choice not just to navigate to the site in question, but to request specific results from the site, knowing that the ads would be there.

IMO, that's quite different from pretty much all other advertising:  I'm not giving 'implicit permission' to view dildo or BMW ads when I go to news sites, or when I click around at random places on the internet.  I don't click on an interesting link because I have an explicit, well defined target.  When I'm browsing, it's like I'm walking in a park seeing the sights, or walking around Burning Man looking at the art.  My expectation is that people won't bother me and that I won't constantly have ads shoved in my face.

Instead, I have to keep up both adblock and a hostban list just so I won't be bombarded with unrequested solicitations. I'm explicitly opting out, yet advertisers continue to try to find ways to bypass that, in spite of my explicitly stated preferences.

I think it would be possible to greatly clean up the ecosystem by effectively banning 'push mode' advertising, and strictly only allowing pull modes such as my two examples above.  Sure, it would mean a hell of a lot less advertising, and a LOT of companies would have to find new ways to survive.  But those companies (and people) will figure something out.  Some of them might even succeed because they have excellent products that people actually want, instead of due to huge propaganda budgets.

Comment by Dentin on Why I Work on Ads · 2021-05-06T11:03:16.841Z · LW · GW

Here's another version of your example:  Some people aren't watching nearly enough snuff and torture videos. There are people who would like to watch them, but don't know it exists.  If I place ads for torture and snuff videos and some people decide to click on them while other people don't, is that a problem?

As I mentioned earlier, advertising is like weaponry.  Your example also reads to me like a classic justification for 'everyone having guns':  "but what if I'm attacked by a rabid dog?  If I have my gun I can protect myself!  See, guns are ok to have!"  Just because it's possible to point out a positive use case, doesn't mean that the remainder of the field is also positive.

And to be clear, I consider your example to be about as likely as the rabid dog example.  Sure, in a world with perfect targeting it could be done, but we're not in that perfect world, and consumers have a vested interest in keeping it that way.  The new privacy initiatives are a big part of that.

Comment by Dentin on Why I Work on Ads · 2021-05-03T22:36:05.085Z · LW · GW

As someone who also works on Ads at Google, I have to take the opposite stance; I view advertising as a blight upon the face of humanity, something to destroy if we can at all figure out how to do so.  I comfort myself knowing that Google Ads is arguably the best of what's an awful ecosystem, and that I work in what's arguably the 'least bad part of advertising', which is fraud and abuse protection.  At least the systems I work on make things less terrible.

However, the 'least bad part of advertising' is still not 'good'.

My favorite analogy for advertising right now is weaponry; specifically, guns.  Advertising is like a handgun. Sure, it can be used for good, and sure, in the right hands it's fine, safe even.  However, the default for a handgun is that it is Unsafe, and you have to put forth effort to "use it for good" because it's entire purpose is to kill living things.  That's advertising - its entire purpose is to alter people's mental state without their permission.  Sure, you can "use it for good", and sure, you can make it 'safe'.  But it's a lot easier to use it for abuse and clockwork orange scenarios.

I'll be switching teams in the next few months to be out of Ads.  Hopefully I can find something positive to work on.

Comment by Dentin on Best empirical evidence on better than SP500 investment returns? · 2021-04-25T10:55:45.156Z · LW · GW

I still have some remaining bitcoin, from the olden days when mortal man could mine it themselves.  My advice to everyone I've ever talked to regarding bitcoin is to avoid it.  I have been slowly divesting my holdings.

My rationale is that while both the dollar and bitcoin are fiat currencies, bitcoin is far, far less anchored to reality than most 'normal' currencies.  The dollar and the euro at least have people trying to keep monetary levels somewhat tied to physical economic value.  The value of bitcoin, meanwhile is largely driven by three things:

  1. propaganda / marketing / bubble behaviour
  2. money laundering
  3. a belief that using bitcoin for transactions has the potential to be cheaper than transactions in the normal financial ecosystem

#1 is basically fad investing.  It can yield huge returns, but is equally likely to yield losses and ultimately just moves money from people who are bad at predicting fads to those who are less bad at predicting fads.

Basing your holdings on #2 is going to be subject to diminishing returns over time, as governments get cranky about it and find ways to crack down.

Basing your holdings on #3 ignores how expensive bitcoin transactions have become, and how you have to either build what amounts to a miniature "normal financial system" on top of it in order to arbitrage the cost, or only do transactions that are sufficiently large that the costs aren't important.

Comment by Dentin on On Sleep Procrastination: Going To Bed At A Reasonable Hour · 2021-04-17T17:58:25.552Z · LW · GW

I've found lighting, melatonin, and caffiene regulation to be wonderful additions to my sleep regime.  I take melatonin pretty consistently at around 8:30 pm, and it seems like it helps make me sleepy ~45 minutes later.  As per SSC though, melatonin isn't particularly strong and the effect I'm noticing may very well be placebo.  I always have caffiene, but rarely after 2 pm, and typically not more than two cups of coffee per day.

That said, I suspect my lighting and light policy is having a much, much bigger effect.

The primary light source in both the computer room and bedroom are a variant of these:

I have a number of presets configured; every single one of them is orange, with no blue enabled, in both rooms.  Even at full brightness, it's rather dim compared to even a single overhead bulb, and I switch to progressively more dim presets later on at night.  I only use the incredibly bright overhead lights if I'm searching for something, or working on a specific high-detail project.

My bedroom has blackout sheets on the windows.

I take all my showers/baths in the dark.  I started doing this to better understand some of my visually impaired friends; I quickly discovered that I liked it more than having the lights on all the time.

All of my computer monitors are all set for the minimum brightness level that still allows reasonable contrast and visibility, as well as the lowest available color temperature.

And lastly, all apps/terminals that I use are set to a black background, except for the browser; my X session background is solid black.  Websites in the browser are set to dark mode css if it's available, and I have a 'dark mode' extension for sites without a dark mode css.  All this said, I generally only flip on the extension in the evening.

I pretty consistently go to bed around 9:30 when I feel super sleepy, and can typically fall asleep in under five minutes.  I've found that even a few minutes of using the overhead lights after 7:00 pm breaks this; I both don't get as sleepy as normal, and find it harder to go to sleep if I go to bed anyway. Caffiene late in the day can also screw it up pretty badly, but that's far more rare than needing the overhead lights for a few minutes for some task.

Comment by Dentin on Goldfish Reading · 2021-03-31T12:21:29.117Z · LW · GW

Quick comment:  I noticed that in all of your examples above, I chunk substantially bigger and fewer pieces.  For example, in the "15 different bold bits" clip, I chunk it into about 8 pieces instead.

This is likely experience/background dependent; I happen to have a relatively strong background in ML and have read a stack of research papers recently, so I probably have both stronger noise filters and more complicated primitives available.

One possibly interesting side note:  I never once, in any of your examples, considered metadata about the topic relevant.  This includes things like the author names, "tested", "study proposed", etc.  I suspect I've learned that 1) author names are almost never important, 2) test procedures are only worth thinking about if they're very explicitly detailed (which was not the case above), and 3) even if the test procedures are ok, they're typically only relevant as a cleanup/sanitization pass once the main concept is understood.

Comment by Dentin on How Should We Respond to Cade Metz? · 2021-02-13T16:47:54.046Z · LW · GW

Let's be blunt here:  the NYT article is pure, unbridled outrage bait dressed up as journalism.  It's not trying to solve a problem, and it doesn't have any agenda other than to pack as much outrage as possible into the publication form factor so as to maximize eyeballs.  It simultaneously craps on EA, the tech industry, SSC, rationalists, MIRI, tech investors and a stack of others.  (I'm surprised that they didn't also include jordan peterson, because hey, why not?)  That's not the sign of someone being honest.

IMO the correct response here is to recommend that friends and family unsubscribe or avoid the NYT. As far as as creating/finding a rebuttal and explaining things to others, don't.  Instead, say the article was a hit piece designed to make everyone look bad, and shrug.  Give it the kind of attention you give to crazy preachers on street corners.  Let it fade into obscurity.

Remember that with outrage bait, you being outraged and complaining about the article to others is entirely the point.  The only winning move is not to play.

Comment by Dentin on Speedrunning my Morning Makes the Coffee Taste Weird · 2021-02-11T22:15:43.336Z · LW · GW

I do a low-grade speedrun in the morning, every day.  If you make it a habit, it becomes less of a stressful "speedrun", and more of "how you do things".


  • Roll out of bed, grab phones
  • While walking through hallway flip on heat
  • Wander to office, put phones on desk where they'll sit all day long
  • Put on clothes and socks that were put on my chair the night before
  • Bathroom
  • Stumble to kitchen
  • Fill teapot, put on stove
  • Fill water purifier back up
  • Put coffee grounds in mug
  • Do a set of pushups
  • Go to office, power up monitors, start catching up
  • Go to kitchen after a few minutes, shut off stove, pour coffee

You know, from the outside, that looks pretty ridiculous.  It is fast and efficient though.  Thank you, Covid lockdown?

Comment by Dentin on We got what's needed for COVID-19 vaccination completely wrong · 2021-02-10T12:33:27.218Z · LW · GW

I think this might be (very slightly) unfair to mRNA vaccines, as the comparison between them and peptide vaccines is pretty situation dependent:

  • We have reason to believe that peptide vaccines will work particularly well here, because we're targeting a respiratory infection, and the peptide vaccine delivery mechanism targets respiratory tissue instead of blood.
  • That said, mRNA vaccines are expected to elicit much stronger immune responses than peptide vaccines would.
  • I suspect that the low efficacy of mRNA vaccines (only 95% - low is relative) is likely because they're only targeting the spike protein, which apparently has 'high mutant escape potential'.  We have a LOT more information about the virus now than we did when the mRNA designs were finalized.  If companies had been allowed to make mRNA vaccine updates without resetting all the clinical trials, I believe we'd have a substantially better/stronger vaccine than just 95%, with a single shot instead of two.
  • mRNA vaccines are really just a technology demo at this point; sure, you can make vaccines with the tech, but that's not the real superpower here.  The superpower is that we have a platform we can use to generate arbitrary proteins in live cells, including proteins we know about but for which we have no reasonable delivery mechanism.  Not all proteins/peptides are water soluble, and fewer are able to get through cell walls.
  • As a technology demo (and not a simple one at that), it's surprising that companies have as much production capability as they do.  I would expect that capacity to be ramped up substantially in the next few years as we start aggressively making use of mRNA treatments to address a host of issues.

To sum up, my view is that mRNA technology is pretty great and I'm really, really glad someone was able to make use of the disaster that was 2021 to ram through safety and clinical trials. My issue is less with the technology and it's large promise, and more with how vaccine testing and rollout has been botched or unnecessarily slowed down at every level.

Peptide vaccines are easy and fast to both design and construct, while being safe unless you really screw up, and effective as long as designed correctly.  However, they do have rather substantial limitations, and it's just happenstance that they're particularly well suited for the situation.  I could see mRNA vaccines having a much wider berth.

Comment by Dentin on Making Vaccine · 2021-02-10T00:18:40.545Z · LW · GW

In my case, I'd estimate that I've spent around two hundred hours over the last several months coming sufficiently up to speed on the topics that I can reason about them.  I started with about your level of biology (or possibly less), but probably a slightly stronger chemistry background.

For the basics, I started with cell biochemistry, DNA/RNA, mRNA and protein construction.  From the vaccine side of things, I just started looking up things I found in the whitepaper which I didn't understand, and once I understood all the terms I started looking for and reading research papers. When I found something I wasn't sure about, I researched it and learned about it.

As examples, in early January, I spent about ten days reading up on VED (vaccine enhanced disease).  Shortly after, I spent a few days digging into chitosan, and trying to understand how sensitive nanoparticle creation is to changes in the mixing process (hint:  not very.) Everything I searched for I was able to find, and pretty much everything reinforced the same internally consistent view of the world.

When you find something that doesn't make sense and you're stuck, write it down, file it away and come back to it later.  Eventually you'll be able to make sense of it.

When you're able to read through most or all of the whitepaper and understand both what's being discussed and why specific things were selected, you'll be in pretty good shape.

It's not particularly difficult, it just takes time and effort.

Comment by Dentin on How do you optimize productivity with respect to your menstrual cycle? · 2021-02-08T11:51:07.557Z · LW · GW

I have two data points for dealing with this successfully, and both amount to "make it stop" instead of "improve things":

  • One of my sisters "doesn't have time for this crap" and one day just stopped taking the placebo pills from her normal birth control and stayed on the active hormones continuously.  This apparently suppressed her period.  After doing this for about ten years, she stopped and successfully conceived at almost 40 years old.  This is not medical advice, do your own research, etc blah blah.
  • One of my female friends has had an IUD for ~15 years now, which over time reduced her period to occasional light spotting and dramatically reduced general symptoms and inconvenience.  The startup period (first year) was rough though, as was the first six months after the first replacement five years in.  The second replacement was basically not a problem.

It should be noted that I'm aware of multiple attempts to use IUDs within friends and family, but only have one data point that was long term successful.  Apparently the first year of an IUD can be brutal and it's not uncommon to give up after a few months.

Comment by Dentin on Making Vaccine · 2021-02-06T00:33:14.473Z · LW · GW

Ugh.  Thanks for the link.

Comment by Dentin on Making Vaccine · 2021-02-06T00:22:12.368Z · LW · GW

I believe that initial post is what got me going down the rabbit hole of peptides and proteins and dna and rna and transcription factors oh my!  It's been a long ride.

Comment by Dentin on Making Vaccine · 2021-02-06T00:20:26.456Z · LW · GW

Sarah Constantin is confused, and likely has not spent significant time reviewing the vaccine design.  From page 32 of the whitepaper:

"Empirical evidence should dominate selection criteria. Here are some best types of evidence:

  • Mapping of epitopes in blood and other samples collected from convalescent patients (ideally stratified by severity of illness). This can be accomplished by a few primary means:
    • 3D structural studies and modeling of neutralizing antibody binding to a viral antigen (e.g. Spike protein)
    • Mapping of linear B-cell epitopes by binding antibodies in convalescent sera to a library of peptides representing viral antigens. A strong signal in a linear epitope mapping study does not guarantee that the epitope peptide in the context of a vaccine will trigger the production of an antibody that binds to this epitope within the context of the virus. However, it is a good indicator that this is at least possible. Peptides can be constrained to approximate native conformation, making it more likely to bind the native epitope.
    • Mapping of T-cell epitopes by stimulating convalescent T-cells with epitope peptides, and measuring their response (e.g. cytokine secretion; ELISpot)
  • Epitope peptides from a peptide vaccine that has shown protection against
  • Successful use of epitope peptides in vaccines that elicit antibodies (or serum)
    effective in virus neutralization assays. B-cell epitopes that allow antibody binding
    to the virus but don’t block viral function might increase risk of
    antibody-dependent enhancement.
  • Mapped epitopes that are effective in virus neutralization assays (e.g. peptides
    compete with viral sequences in cellular infection assays).
  • Successful use of epitope peptides in vaccines that elicit T-cell responses, or
    peptides shown to stimulate T-cells or cytokine production in ELISpot or other
    T-cell assay in cells from convalescents."
Comment by Dentin on Making Vaccine · 2021-02-06T00:15:05.107Z · LW · GW

Sorry about that; I believe I misread your comment as implying that if the moderator is ignorant, he won't have enough information to form a reasonable prior.  My disagreement was along that line, as it seems that misinformation, especially about medical things, is so prevalent that everyone's default prior should be 'fraud unless lots of evidence points the other way'.

Comment by Dentin on [Link] Sarah Constantin on RaDVaC · 2021-02-05T23:20:10.008Z · LW · GW

A couple of minor quibbles:

  1. The peptides did not "come from in silico studies"; they came from the antibody profiles of real patients, who really had covid, and really recovered from covid (ideally without having a Bad Time during recovery.)  So there's more than just computational reasons to believe they will be useful.
  2. It's questionable to complain that radvac did not have a single "research study using any of the peptides in the RADVAC white paper that found they inhibited SARS-CoV-2 infection in cells, let alone animals or humans", when the pfizer vaccine (which we know works) was designed via the same process a year ago, when we had even less information about effectiveness.
Comment by Dentin on Making Vaccine · 2021-02-05T12:31:50.216Z · LW · GW

Yes.  The differential tradeoff is how one should evaluate this.  The only reason my evaluation came out in favor of trying the radvac vaccine is because I have a high-risk event coming up in the next few months, and I am extremely unlikely to be able to acquire a commercial vaccine before then.

Comment by Dentin on Making Vaccine · 2021-02-05T12:28:27.107Z · LW · GW

In my case, yes.  My bio expert indicated that it was likely to be effective (more than 50%, but less than 90%) and that the risks were effectively zero in terms of serious complications.

Regarding the food grade versus lab grade question, as well as inaccuracies or mistakes in construction of the vaccine, this was a question I spent a reasonable amount of time on.  The TL/DR is that the engineering tolerances are incredibly wide; the molecular weight of the chitosan isn't that important, the mixing rate isn't that important other than it be fast enough, the quantities aren't that important, exact peptide quantities aren't that important etc.  A lot of these can be off by not just percentage points, but integer factors, and the result will still be acceptable.

It's also worth pointing out that unless you make serious, significant mistakes that dramatically impair effectiveness, you can always just use "more dakka" to overpower the variations.  My plan is to mix each batch independently, such that at least some of the construction variations are expected to cancel.  (Also, freezing the final vaccine is likely to impair effectiveness, from what little I've found on the topic.)

Comment by Dentin on Making Vaccine · 2021-02-05T12:20:37.533Z · LW · GW

Again, I have to disagree - misinformation is much more likely than information by default, and the moderator need only have a reasonable low-probability prior in order to reject unusual/uncommon claims without evidence.

Comment by Dentin on Making Vaccine · 2021-02-05T12:13:37.073Z · LW · GW

Yeah, the pfizer vaccine looks like it just uses mRNA to construct the RBD (receptor binding domain) of the spike protein, which is about two hundred amino acids long.  None of the default 9 peptides in gen 9 radvac are for that domain.  See page 40 of the whitepaper for the full spike protein sequence; the highlighted blue is the RBD, and the short underlined sequences are peptides selected for the vaccine.

The moderna vaccine uses mRNA to construct pretty much the whole spike protein, including the RBD.  This has overlap with 3 of the 9 radvac peptides.

This paper has one of the better lists I've found of commercial vaccine types:

From the list in that paper, it seems like pretty much all commercial products are using spike as the primary target; radvac is unique in that it also targets ORF and Nuc.


Amusingly enough, when talking about the commonly targeted spike protein RBD sequences, the radvac whitepaper lists on page 29:  "Spike 450-500; ACE2 binding residues of the RBD (Zhang et al); low degree of conservation; probably moderate to high mutant escape potential".  So they pretty much called it in regards to the new virus strains with mutations like E484K.

Another radvac whitepaper quote, which seems to line up with my independent research:  "It is important to note that most published neutralizing antibodies target Spike RBD, as do many vaccines in commercial development. However, given the high degree of mutability of the RBD portion of Spike, it is highly recommended to identify and select targets outside the RBD because of mutant escape potential."

And lastly, more specific to why no RBD peptides were selected, page 34:

"Therefore, rather than focusing on ACE2-binding epitopes in the highly mutation prone RBD to inhibit virus binding to the ACE2 receptor, we targeted these B-cell epitopes in the highly conserved portions of the Spike protein to strategically neutralize proteolytic cleavage and membrane fusion. Furthermore, all three are bound by antibodies present in the sera of large fractions of convalescents, and they produce among the highest signals in linear epitope mapping studies, which are far higher than signals measured for binding to any linear epitope in the RBD."

Comment by Dentin on Making Vaccine · 2021-02-05T11:53:36.298Z · LW · GW

Absolutely obviously yes.  I have some level of concern that this post will go viral (ha ha), get a lot of attention outside of lesswrong, and the company I'm working with will cancel my order because it's "covid misinformation" related.

The FDA might be slow and take months to approve safe things while thousands of people die per day, but they're perfectly capable of announcing an immediate and indefinite peptide ban in under a day because a news article crossed the wrong person's desk.

Comment by Dentin on Making Vaccine · 2021-02-05T11:44:53.732Z · LW · GW

My estimate for whether or not I would test positive on a blood test was only about 50%, since blood isn't the primary place that the response is generated.  I'm already betting a substantial amount of money (peptide purchases and equipment) that this will be helpful, and I see no reason to throw an additional $50 on a break-even bet here.

I would, however, be happy to commit to sharing results, whether they be positive or negative.

... and now it occurs to me that if Lesswrong had a 'public precommitments' feature, I would totally use it.

Comment by Dentin on Making Vaccine · 2021-02-05T11:36:12.435Z · LW · GW

A lot of people have been working really hard for the last year to discover, understand, and know these things.  It's the foundation for how the mRNA vaccines work.

Perhaps take a look through this:

Comment by Dentin on Making Vaccine · 2021-02-04T23:31:04.073Z · LW · GW

While I generally agree with the concept, I'm going to push back a little here.  I read the 1-2% chance as less being about "why aren't companies doing it" and more about lack of information.

My initial reaction to seeing it was that it was a combination factors along the lines of:

  • "there's a lot of fraud out there, and by default my prior for things like this being valid is very low"
  • "factoring in that a couple of lesswrongers seem to think it's ok that only pushes my estimate up into the handful of percent range"
  • "but there's also evidence against, in that we don't see any commercial products based on this, which pushes my estimate down to 1-2%"

I think this is a pretty reasonable place to start from.

Comment by Dentin on Making Vaccine · 2021-02-04T22:50:03.394Z · LW · GW

Yes, I still plan to get the commercial vaccine once it's available to me (likely some time in august.)  As I understand it, the commercial vaccines hit different areas of the virus from the ones that radvac selected, improving protection even further.

There is actually an optional peptide for radvac which does cover one of the same regions as the commercial vaccines.  I elected not to include it under the assumption I'd be getting it from the commercial vaccine.

Comment by Dentin on Making Vaccine · 2021-02-04T22:37:23.093Z · LW · GW

Yes; sorry I was unclear.  Those peptides generate the antibodies we care about, that are known to be effective against the full virus.

Comment by Dentin on Making Vaccine · 2021-02-04T12:50:41.166Z · LW · GW

Regarding the 33.4% approval rate:  based on what I've learned about traditional vaccine development and production in the last few months, I am not at all surprised.  Both peptide and RNA vaccines are effectively "state of the art" technologies compared to traditional vaccine techniques.  It's like comparing modern non-invasive out-patient surgery to the 1970's equivalent.

You need look no further than the russian and chinese vaccines - those use the rather crude technology of "throw big chunks of inactivated virus particles at the immune system and hope that the immune system guesses the right antibodies to deal with the live version."

Both peptide and RNA vaccines are instead, "we have identified very specific antibodies which we know are effective both from the serum of recovered patients and from computational modeling, then use exactly the minimal protein sequences needed to generate those antibodies."

Both the russian and chinese vaccines use chunks of proteins that are thousands (and likely tens of thousands) of amino acids long, in a mostly inactivated form.  The immune system has no idea what to latch onto, what will be effective at stopping replication, but it does generate a bunch of randomish antibodies anyway just in case.  In a lot of people, this is enough to take the edge off getting stick.

The peptide and RNA vaccines on the other hand, are extremely narrow.  The radvac vaccine for example targets 9 specific virus protein sequences, the shortest of which is 10 amino acids long, and the longest of which is 25.  Each of these sequences is from an empirically likely effective antibody, found in real people who have really recovered.

Like I said, very, very different technologies.

Comment by Dentin on Making Vaccine · 2021-02-04T12:34:38.182Z · LW · GW

My understanding is that it helps a lot.

The biggest benefit seems to be that the immune system is primed in at least some fashion; it knows what to look for, and it has readily available tools that should be effective.  It doesn't have to take a day or a week to try random things before it finally discovers a particularly effective antibody and gets the production chain ramped up to start a proper immune response.

Instead, your immune system will very quickly get a signal it understands as bad and can immediately start ramping up when it does detect the virus.

Keep in mind that the commercial vaccines don't have 100% success rate in that some people still get sick, but the 'priming' of the immune response is still there.  I believe this is why the death rate / severe complications rate is effectively zero for immunized patients, even though it's possible to get sick.

(Again, my understanding.  I would very much appreciate correction/clarifications here.)

Comment by Dentin on Making Vaccine · 2021-02-04T12:27:45.849Z · LW · GW

I agree as well.  It takes a non-trivial amount of knowledge and research to evaluate the the whitepaper and its claims, and I wouldn't expect the moderator of a "neoliberal" group to have that expertise.  We have options with a known risk profile (the commercial vaccines), and there's a spot of fraudulent "cures" out there.  The safe thing for a moderator to do is blackhole potentially dangerous claims they don't have the time and/or experience to evaluate.

Comment by Dentin on Making Vaccine · 2021-02-04T12:24:16.767Z · LW · GW

Regarding the final paragraph, "you need some level of expertise yourself before you can distinguish real experts from fake":  that has been the number one reason I didn't beat johnswentworth to the punch and post first with my experience.

I have learned more about biochemistry in the last three months than in my entire prior life combined.  It has taken me three months of research, asking questions, and conferring with experts to get sufficient confidence in my understanding to commit to the project.

I'm incredibly thankful to you (johnswentworth) for posting this article; it tracks almost perfectly with my understanding, and I have no significant model conflicts with any of your observations.  It raises my confidence in both my understanding, and the project, substantially.

Comment by Dentin on Making Vaccine · 2021-02-04T12:18:16.224Z · LW · GW

My personal estimate is that the the percentage of nurses who have done this is effectively zero (less than one in a thousand with high probability, less than one in ten thousand with moderate probability.)

Further, those who did do it are likely to have read through the whitepaper, and therefore are also likely to get the commercial vaccine, as it covers different epitopes than the radvac vaccine.

Comment by Dentin on Making Vaccine · 2021-02-04T12:15:19.235Z · LW · GW

The whitepaper is a good source, but like johnswentworth, I also contacted a medical professional to evaluate it.  The response came back quickly and confidently, and was along these lines:

"Oh, yeah, this is safe.  Nasal vaccines are safe.  The biggest worry is that it might not work, so make sure you get the commercial vaccine too.  I'd be interested in doing this with you as a joint project and giving it to my family, and I also have a colleague who might be interested in doing it."

The biggest point of disbelief on their part was that it's possible to order all the equipment and peptides online and have them shipped to your door.

Comment by Dentin on Making Vaccine · 2021-02-04T12:07:44.928Z · LW · GW

My rough guess is that there's a 75% probability of effectively full immunity, and a 90% probability of severity reduction.  This is a pretty well tested and understood vaccine mechanism, and the goal isn't "perfect immunity" as "prime the immune system so it doesn't spend a week guessing about what antibodies it needs to combat the virus effectively".

As to why established companies don't do it, I believe it's partially logistics, and largely red tape. Logisitics first (though it should be noted that at least some of these could likely be tackled with a bit of effort):

  • Shots are well understood and easy; people are used to them, people know how to give them, etc.  Nasal spray is irritating and makes you want to blow your nose, which washes out a lot of it and reduces effectiveness.
  • You need multiple of these annoying doses in the nose, staggered a few days apart, to generate a 'good' response.
  • This particular nanoparticle vaccine doesn't have a long shelf life due to peptide degradation.  Peptides don't last forever, and while they're more stable than the RNA vaccines, you'd have to ship them frozen as well.
  • Nanoparticle vaccines in general suffer from particle aggregation over time.  The particles will gradually aggregate in solution, and if you freeze it they aggregate faster; freezing changes the size distribution pretty dramatically.  That said, I don't know how much that impacts effectiveness, and it doesn't seem like an extensively researched topic.  I only found one paper discussing it.
  • Because it doesn't have a long shelf life, it has to be mixed in-house, then distributed, preferably within a small number of days.  I plan to mix a new batch every week for my prime and boosters.  On the plus side, mixing it can actually be done at home with pretty cheap and ordinary tools, it just takes time.

Red tape on the other hand is a huge problem and flat out intractable.  Unlike injections, companies need to get safety approvals and testing done for pretty much all the components and the delivery mechanism, not just the active ingredients (the peptides.)  Then, separate approvals and testing for each individual set of peptides, via a process that operates closer to the scale of decades than months and costs billions of dollars.  Any modification to the peptide set can be expected to restart the process from scratch.  Then, companies also need to set up / reconfigure their legal strategy to protect themselves, and ensure that things are sufficiently balanced business-wise that they don't go bankrupt.

Using ancient, primitive "stab people in the arm" technologies isn't great, but it erases likely more than half of the regulatory burden, because "we've been doing that for a century and we know it's only somewhat dangerous", wheras new technologies like "snort some nanoparticles" are Scary and Dangerous and New and Have Not Yet Been Approved By Appropriately Serious People Being Serious.

Nevermind that every time you play with a pet, breathe in part of a cloud of dust, or smell your SO's hair, you're inhaling more foreign peptides than what's present in a vaccine dose.

Comment by Dentin on Making Vaccine · 2021-02-04T11:42:40.182Z · LW · GW

Oh, it's far, far worse (better?) than $2 per dose.  As a thought experiment, I price estimated buying enough peptide for a hundred thousand doses, and it only costs about ten thousand dollars.  Ten cents a dose is closer to realistic if you buy in bulk.

Which also brings to mind a question of civilizational inadequacy:  if we really cared and it really mattered, why not have every university with a lab in the country crank out a hundred thousand doses per week to their local populace?

Comment by Dentin on Making Vaccine · 2021-02-03T21:54:05.369Z · LW · GW

Speak of the devil.  I literally just placed my peptide order a couple of hours ago.  My experience (finding supplies, test runs of mixing the solution, safety profile, analysis, etc.) basically matches up with this post.


Thanks a lot for posting it.

Comment by Dentin on Grokking illusionism · 2021-01-06T23:18:46.504Z · LW · GW

I don't at all share that intuition.  My intuition is that consciousness being emergent is both extremely plausible, and increasingly likely as system complexity increases.  This intuition also makes consciousness approximately as "real" as "puppies" and "social media".  All three are emergent phenomena, arising from a very small and basic set of primitives.